Evaluation of In-Car Communication Systems - Digital Signal Processing for In-Vehicle Systems and Safety - page 72

with ICC systems usually also have hands-free and speech dialog systems installed. If
these systems operate together, special care has to be taken. For example, if the ICC
system adapts the playback volume in dependence of the background noise level, an
echo cancellation unit that is included in hands-free or speech dialog systems will
have to reconverge. Some details about those restrictions can be found in [ 3 , 32 ].
The connection and interaction with other subsystems of the vehicle also leads to
several advantages that can be exploited beneficially. External signals 2 such as
music, navigation prompts, or warning signals can be used not only to identify the
feedback paths in terms of the critical frequencies (those where howling would
start) or the delay but also to estimate the impulse or frequency responses between
all loudspeakers and all microphones. Furthermore, information that can be
extracted from the automotive bus systems such as the CAN 3 bus also can help to
improve the operation of ICC systems. For example, the results of weight sensors
that are installed in seats in order to warn the passengers about not having fastened
their seat belts can be used to deactivate the ICC processing for seats that are not
occupied. However, going into the details of the interaction between ICC systems
and other subsystems of a car is beyond the scope of this chapter. In the following,
we will give a brief description of those signal processing components that are
necessary for ICC systems.
To improve the stability margin, signal processing such as beamforming, feed-
back and echo cancellation, adaptive notch filtering, noise suppression, adaptive
gain adjustment, equalization, and nonlinear processing can be applied. We will
describe these components briefly in Sect. 5.3 . In this section, we will focus on the
general concept of ICC systems as depicted in Fig. 5.4 .
Even if most automotive signal and data transfer options such as the MOST 4 bus
are able to transport audio signals on a sample-by-sample basis, virtually all audio
signals are transported and processed as blocks of samples, mostly using a minimal
block size of 64 samples in today's cars. For that reason, it is of advantage since
signal enhancement units such as ICC systems use block-based algorithmic
approaches, but it leads, as a side effect, to a lower computational complexity
compared to sample-by-sample processing. As a result, the very left and the very
right signal processing blocks in Fig. 5.5 are transformations such as FFTs or more
sophisticated filter banks that transform a block of signals into the subband domain
and vice versa, respectively. As we will highlight later on, delay is a very critical
issue for ICC systems. As a result, low-delay filter bank approaches are of special
importance here. In addition, some pre- and postprocessing such as preemphasis or
de-emphasis filters are usually performed. More details are described in Sect. 5.3.1 .
After changing the signal representation from the time domain to the subband
(or frequency) domain, some algorithmic components are applied to all input
2 In this case, external is meant to be from the point of view of the ICC system.
3 CAN stands for controller area network.
4 MOST is the abbreviation for media oriented systems transport . This bus system is often used for
the transport of audio data.